[Ffmpeg-devel-irc] ffmpeg.log.20160705

burek burek021 at gmail.com
Wed Jul 6 02:05:02 CEST 2016


[00:43:22 CEST] <jukejuke> probably not the best time to ask, but is there any way to more securely pass http basic auth credentials to ffmpeg?
[00:43:56 CEST] <jukejuke> currently I am just doing 'ffmpeg -i http://user:pass@host.com/file out.mp4' but this exposes the credentials in the process list
[00:45:02 CEST] <neuroserpens> So... I followed the mosaic tutorial and got 3 videos playing at the same time. But they are 3 pieces of a song, so my first problem pertains to what parameters I should include to make each one start at a specific time and the second problem relates to the final output mosaic only having the audio from the first video instead of all 3.
[00:47:04 CEST] <jukejuke> neuroserpens: i believe this is what you need for the audio https://www.ffmpeg.org/ffmpeg-filters.html#amix
[00:50:29 CEST] <neuroserpens> jukejuke: alright... let's see what happens... edited the comand and waiting for the output again
[00:50:58 CEST] <jukejuke> i think I may have misunderstood what you're trying to do actually, you want to create a video with quarters whereby each quarter plays one after the other?
[00:56:12 CEST] <neuroserpens> jukejuke: no
[00:56:24 CEST] <neuroserpens> jukejuke: I need to hear myself playing all instruments at the same time so the song will make sense
[00:57:05 CEST] <neuroserpens> jukejuke: YES!!! it worked!!!!!!
[00:57:10 CEST] <neuroserpens> jukejuke: now i can hear all 3
[00:57:33 CEST] <neuroserpens> jukejuke:  now i need to sync them... and maybe regulate each video's volume if it's possible
[00:57:41 CEST] <jukejuke> great, i think this is the sync solution: https://www.ffmpeg.org/ffmpeg-all.html#setpts_002c-asetpts
[00:59:00 CEST] <neuroserpens> jukejuke: i'm not sure as to how i should use that
[00:59:09 CEST] <neuroserpens> jukejuke: syntax wise
[00:59:25 CEST] <neuroserpens> jukejuke: ah there are examples below... nevermind sorry
[00:59:29 CEST] <neuroserpens> jukejuke: that was dumb of me
[01:00:13 CEST] <ossifrage> Is there a faster alternative to the select filter to just decode i-pictures?
[01:09:10 CEST] <neuroserpens> jukejuke: mother of god this trial and error on the timings is killing me
[01:16:20 CEST] <jukejuke> neuroserpens: cut a fixed amount from each of the sources so the remaining videos are only about 10 seconds each, then work out your offsets and then add the length of the amount you clipped
[01:16:59 CEST] <neuroserpens> jukejuke: i don't mean to be a vampire but how do i cut
[01:20:09 CEST] <jukejuke> neuroserpens: if you do this for each of your sources you should end up with a set of 10 second videos: ffmpeg -i input1.mp4 -t 10 output1.mp4
[01:20:29 CEST] <neuroserpens> jukejuke: very interesting. thank you very much
[01:30:39 CEST] <neuroserpens> jukejuke how do i set each video's volume? one is way louder than the others
[03:32:05 CEST] <MichaelTiebesl> good day, can somebody help me with this: i have almost 40 folders with m4a files in Downloads. I want to convert that to ogg, how to do in terminal in linux?
[03:49:09 CEST] <klaxa> MichaelTiebesl: write a script that iterates through folders and then iterates through files, converts them and that's basically it
[03:49:50 CEST] <klaxa> basically like: for dir in *; do cd $dir; for f in *.m4a; do ffmpeg [...]; done; cd ..; done
[03:52:32 CEST] <MichaelTiebesl> klaxa:sorry but writing script i can't
[03:52:32 CEST] <klaxa> with: ffmpeg -i "$f" -c:a vorbis "${f/m4a/ogg}"
[03:52:45 CEST] <klaxa> i just wrote it in one line for you
[03:53:02 CEST] <klaxa> you can literally copy that into your bash
[03:53:42 CEST] <klaxa> just replace the ffmpeg [...] with the line i posted just now
[03:54:03 CEST] <klaxa> it's not tested though
[03:54:14 CEST] <MichaelTiebesl> so i make a bash script and but this line in? ffmpeg -i "$f" -c:a vorbis "${f/m4a/ogg}"
[03:54:22 CEST] <klaxa> yes
[03:54:45 CEST] <klaxa> if the first for loop iterates through files instead of directories it will break
[03:55:01 CEST] <MichaelTiebesl> klaxa:and what if i want a good quality, i saw on google it works from 1 to 10
[03:55:02 CEST] <klaxa> so are there also other files in the directory where your 40 or so directories are?
[03:55:31 CEST] <MichaelTiebesl> in the folders are jpg files and mp3 files
[03:55:46 CEST] <klaxa> that is not a problem
[03:56:07 CEST] <klaxa> within the directories it will only match m4a files
[03:56:58 CEST] <klaxa> i'm not sure if -q:a 10 will set the vorbis quality to 10
[03:57:13 CEST] <MichaelTiebesl> how do i call that document in my home folder?
[03:57:16 CEST] <MichaelTiebesl> any name?
[03:58:20 CEST] <klaxa> encode_m4a_to_ogg.sh ?
[03:58:26 CEST] <klaxa> dunno
[03:59:22 CEST] <furq> it looks like -q:a maps to libvorbis -q
[03:59:26 CEST] <furq> except for negative values
[04:00:02 CEST] <MichaelTiebesl> like this http://pastebin.com/gF5Djimd
[04:00:17 CEST] <klaxa> no
[04:01:07 CEST] <klaxa> MichaelTiebesl: like this: http://pastebin.com/NuitNaeP
[04:01:16 CEST] <furq> MichaelTiebesl: find . -name "*.m4a" -exec ffmpeg -i "{}" -vn -q:a 8 "{}".ogg \;
[04:01:22 CEST] <klaxa> or that
[04:01:37 CEST] <furq> there's no need to put it in a script unless you want to reuse this and you can't remember the command
[04:01:46 CEST] <furq> and you don't bother saving your bash history
[04:01:55 CEST] <klaxa> i stopped trusting my bash history
[04:02:08 CEST] <klaxa> with multiple sessions it sometimes messes up
[04:02:16 CEST] <klaxa> and i lose a lot of commands i used
[04:02:17 CEST] <furq> there are ways around that
[04:02:22 CEST] <furq> they're a bit hacky though
[04:03:18 CEST] <furq> maybe putting it in a script will be more helpful if you don't run find -exec several times a week like i do
[04:03:31 CEST] <furq> but you should, because it's great
[04:03:57 CEST] <klaxa> i have to remember that syntax, it looks quite concise
[04:05:20 CEST] <MichaelTiebesl> thanks both
[04:05:23 CEST] <furq> it's annoying that you can't do parameter expansion to get rid of the file extension
[04:05:34 CEST] <MichaelTiebesl> im looking at it right now but got confused
[04:05:40 CEST] <furq> maybe you could do that with xargs but i don't use that often enough
[04:06:08 CEST] <MichaelTiebesl> furq:has it with the quality in his script i see
[04:07:27 CEST] <MichaelTiebesl> can i do it like this: http://pastebin.com/u68ciZbA
[04:07:33 CEST] <MichaelTiebesl> combo from both\
[04:07:48 CEST] <furq> the default is -q:a 3 by the looks of things, which is pretty bad
[04:08:07 CEST] <furq> also you don't need done at the end
[04:08:33 CEST] <furq> and it should be "#!/bin/sh" if you're not using any bash features (which you're not)
[04:09:05 CEST] <furq> otherwise this script won't work if you decide to install freebsd, which is obviously a very likely thing to happen
[04:12:24 CEST] <MichaelTiebesl> do i put chmod +x first?
[04:12:46 CEST] <furq> sure
[04:13:31 CEST] <MichaelTiebesl> hmm i think i have it correct then
[04:14:13 CEST] <MichaelTiebesl> furq:thank i see i have convert a folder so far
[04:14:40 CEST] <MichaelTiebesl> what if i want mp3 to ogg? i just change the m4a in mp3?
[04:14:52 CEST] <furq> right
[04:15:32 CEST] <MichaelTiebesl> thanks really appriciate it
[04:16:36 CEST] <MichaelTiebesl> i convert 1 m4a file into a ogg file but m4a is still in the name i see, like: 01 Dedication.m4a is now 01 Dedication.m4a.ogg
[04:40:34 CEST] <MichaelTiebesl> furq:still there?
[04:41:54 CEST] <furq> hi
[04:43:28 CEST] <MichaelTiebesl> furq:i convert 1 m4a file into a ogg file but m4a is still in the name i see, like: 01 Dedication.m4a is now 01 Dedication.m4a.ogg
[04:43:33 CEST] <MichaelTiebesl> anu idea?
[04:43:39 CEST] <MichaelTiebesl> any*
[04:43:45 CEST] <furq> there's no way around that if you use find
[04:45:47 CEST] <MichaelTiebesl> okay then i have to rename it by command then after
[04:47:27 CEST] <furq> shopt -s globstar; for f in **/*.m4a; do ffmpeg -i "$f" -vn -q:a 8 "${f%.*}.ogg"; done
[04:54:11 CEST] <MichaelTiebesl> thanks will try that
[04:59:16 CEST] <MichaelTiebesl> furq:wow thanks for that. im so happy with that
[05:01:52 CEST] <MichaelTiebesl> does it make sense to convert 256 kbps to 320kbps
[05:02:05 CEST] <MichaelTiebesl> it is 256kpbs now
[05:02:12 CEST] <MichaelTiebesl> or i leave it like this?
[05:02:19 CEST] <furq> it makes as much sense as any other lossy conversion
[05:02:26 CEST] <furq> -q:a 8 is nominally 256kbps anyway
[05:05:10 CEST] <MichaelTiebesl> furq:i leave it with 256K im happy with what you did :)
[05:42:56 CEST] <ossifrage> Is there an video filter option for a higher quality image reduction?
[05:43:48 CEST] <Spring> ossifrage, there are different scaler algorithms that can be set
[05:43:53 CEST] <ossifrage> I replaced 'scale=240:-1' with 'scale=960:-1,scale=480:-1,scale=240:-1' and the result looked quite a bit better
[05:44:16 CEST] <ossifrage> Sprint where is that documented?
[05:44:21 CEST] <Spring> https://ffmpeg.org/ffmpeg-scaler.html
[05:44:36 CEST] <furq> you can also use -vf zscale if your build has it
[05:45:06 CEST] <furq> -sws_flags spline should give good results with swscale
[05:46:35 CEST] <Spring> would be amazing if the type of neural network scaling waifu2x can do was fast enough for video
[05:52:26 CEST] <ossifrage> Ah, that looks much better and is faster then my command line pyramid scaler
[05:59:25 CEST] <ossifrage> Way back in a previous life I did a bunch of encoding work using 'A New Hope' as test footage, now I messing with encoding stuff again and I'm using 'The Force Awakens'
[06:00:18 CEST] <ossifrage> I wish I could find about 10-15 minutes of high quality source material, I converted a bunch of red raw footage I found but that is all short clips
[06:06:58 CEST] <Spring> ossifrage, no Blu-Rays on hand?
[06:07:31 CEST] <ossifrage> That is what I'm using a blu-ray of The Force Awakens
[06:11:13 CEST] <ossifrage> I can't tell you how much off by one errors drive me crazy. The text isn't being drawn based on the baseline, see: http://imgur.com/a/HHQ1f
[06:12:12 CEST] <ossifrage> The 7 8 and 9 are in the same place and then the 10 jumps down by a line
[06:12:16 CEST] <Kirito> Anyone have any recommendations on resources for x264 video encoding? Specifically for fine tuning x264 options instead of just relying on presets (something to provide an overview of the most significant options, in what situations it is good to change them and to what extent, etc.)
[06:15:51 CEST] <furq> Kirito: http://dev.beandog.org/x264_preset_reference.html and https://en.wikibooks.org/wiki/MeGUI/x264_Settings
[06:16:01 CEST] <ossifrage> Ouch this 240x100 video is encoding at 0.85fps with x265, I wonder what I broke
[06:18:51 CEST] <ossifrage> Not only is it going very slowly, it is producing broken video, doh...
[06:19:23 CEST] <Kirito> furq, Thanks! I had used the second link before but couldn't find it again, and I never acually found anything listing all the default settings for individual presets, that's extremely useful
[06:19:54 CEST] <furq> there isn't usually much reason to deviate from the presets
[06:22:00 CEST] <furq> that also doesn't mention tunings, which will make a much bigger visual difference
[06:27:33 CEST] <Kirito> Yeah, I'm using --tune film currently. I've never really taken the time to learn how to fine tune the settings though, so I'm not sure if it is something worth doing.
[06:28:02 CEST] <Kirito> For example though, my input video contains content recorded at night (fireworks), so roughly 70% of the video is almost solid black. I feel like there are probably things you can tune to improve quality and compression when accounting for that
[09:00:39 CEST] <furq> has anyone used h264_omx on an rpi yet
[09:00:48 CEST] <furq> i figured i'd try it out but i can't get it to work here
[09:00:51 CEST] <furq> http://vpaste.net/iO1QY
[09:24:25 CEST] <furq> nvm i figured it out
[10:16:04 CEST] <dinux5> Hello! Which FFmpeg module can help me display a video on multiple displays .
[10:16:57 CEST] <dinux5> Say if there are four screens, then 1/4th of the video frame must be clipped at the server and sent to that particular client
[10:46:09 CEST] <soulshock> with 2-pass x264 encoding, do I need to run the 1st pass for each target bitrate or can I run pass 1 once and use the same resulting stats file for each target bitrate?
[10:50:15 CEST] <furq> soulshock: iirc it should be fine as long as each output is the same resolution
[10:50:34 CEST] <furq> and same gop length, but i assume you wouldn't change that
[10:51:04 CEST] <soulshock> yeah I was planning to use pretty much the same settings for each bitrate, except the bitrate of course
[10:52:43 CEST] <furq> apparently with x264 you can use --pass 3 for the second pass to update the passfile at the same time as reading it
[10:52:50 CEST] <furq> i'm not sure how much benefit that gives or whether it works with ffmpeg
[11:18:31 CEST] <binarym> hey. I noticed that my mp
[11:19:13 CEST] <binarym> so, again... i noticed that, in my mp4 encoded with ffmpeg, i have metadata holding encoder options, like this : core 133 r2334 a3ac64b - H.264/MPEG-4 AVC codec - Copyleft 2003-2013 - http://www.videolan.org/x264.html - options: cabac=1 ref=4 deblo [...]
[11:19:25 CEST] <binarym> is there a way to avoid it ?
[11:20:25 CEST] <binarym> is it the purpose of -map_metadata with an empty set ?
[12:20:04 CEST] <soulshock> furq yeah I see it can do N passes
[12:25:07 CEST] <dlaboratory> tried to encode videos using `hevc_qsv' encoder
[12:25:15 CEST] <dlaboratory> but got this error
[12:25:16 CEST] <dlaboratory> [hevc_qsv @ 0000000002ddefe0] Could not load the requested plugin: 2fca99749fdb49aeb121a5b63ef568f7
[12:25:48 CEST] <dlaboratory> I'm using a Broadwell CPU, which is supposed to have hardware HEVC acceleration
[12:26:04 CEST] <dlaboratory> and when I ffplay -vcodec hevc_qsv, it crashes
[12:33:43 CEST] <jkqxz> dlaboratory:  Broadwell does not have H.265 hardware.  I think there is some GPU acceleration in the Media SDK, but I don't know what the constraints on that are.
[12:34:13 CEST] <dlaboratory> aha, Broadwell doesn't have H.265 hardware?
[12:35:33 CEST] <dlaboratory> That's sad. Now playing a 4K H.265 video using CPU only decoding, unacceptably slow.
[12:35:40 CEST] <jkqxz> No.  Decode was added in Braswell and encode in Skylake.
[12:35:44 CEST] <Fyr> =)
[12:36:10 CEST] <Fyr> I hope, my nvidia has hardware decoding.
[12:36:43 CEST] <dlaboratory> I need to transcode it to H.264 in order to get a smooth playback :(
[12:39:21 CEST] <BtbN> decoding hevc via vaapi should work fine.
[12:43:51 CEST] <dlaboratory> thanks, but I'm using Windows
[12:44:52 CEST] <dlaboratory> the problem is knotty now... I have to either transcode it to H.264 for hardware decoding, or downscale it to 1080p. Both of them take a long period of time on this low voltage CPU...
[12:46:44 CEST] <Fyr> no hardware acceleration even when transcoding?
[12:47:33 CEST] <brontosaurusrex> dlaboratory: what players did you try?
[12:47:51 CEST] <dlaboratory> ffplay, potplayer, and Windows built-in video player
[12:48:04 CEST] <brontosaurusrex> mpv, mpc-hc?
[12:48:11 CEST] <dlaboratory> yes
[12:48:15 CEST] <dlaboratory> also tried mpc-hc
[12:48:51 CEST] <dlaboratory> all of them gave bad experience, because of the poor performance of low voltage mobile CPU
[12:49:16 CEST] <dlaboratory> I'm struggling to find a way to `hardware decode' it
[12:49:38 CEST] <dlaboratory> when invoke ffplay with `-vcodec hevc_qsv', it crashes
[12:49:48 CEST] <Fyr> durandal_1707, have you pushed your ffmpeg-mt extensions?
[12:49:56 CEST] <dlaboratory> with DXVA turned on, potplayer crashes
[12:50:02 CEST] <Fyr> xD
[12:50:06 CEST] <brontosaurusrex> dlaboratory: but hevc 1080p would play?
[12:50:10 CEST] <dlaboratory> the rest just `stopped responding'
[12:50:22 CEST] <dlaboratory> I haven't actually tried
[12:50:36 CEST] <dlaboratory> but I think 1080p HEVC wouldn't be a problem
[12:51:37 CEST] <dlaboratory> I started transcoding it to H.264, preset ultrafast, speed factor: 0.211x :(
[13:07:13 CEST] <brontosaurusrex> dlaboratory: probably the question  would be: what is a very fast downscaler
[13:07:30 CEST] <dlaboratory> yeah
[13:12:02 CEST] <jkqxz> If you want your transcode to be faster you should be able to use the H.264 encode hardware, freeing up the whole CPU to do the decode part of it.
[13:12:35 CEST] <dlaboratory> yes, i'm using h264_qsv
[13:12:59 CEST] <dlaboratory> I think the bottleneck is still 4K HEVC CPU decoding
[13:13:42 CEST] <dlaboratory> when playing the video, the framerate is about 2fps or less, if the video player doesn't crash immediately :P
[13:14:47 CEST] <jkqxz> You are?  I thought qsv doesn't have an ultrafast preset.
[13:17:23 CEST] <dlaboratory> I was originally using CPU h264 encoder
[13:17:43 CEST] <dlaboratory> then right after I restarted using h264_qsv
[13:17:58 CEST] <dlaboratory> speed factor is still about 0.2x
[13:18:15 CEST] <dlaboratory> that suggests the bottleneck is hevc decoding, not h264 encoding
[13:18:38 CEST] <dlaboratory> poor cpu
[14:02:57 CEST] <friki2015> hi, is it correct as an output stream? ffmpeg ... -f flv rtmp://john:mysecret@localhost/livestreaming
[14:03:56 CEST] <friki2015> it seems that ffmpeg parse "john" as the hostname :S
[14:04:37 CEST] <friki2015> according to the documentation it seems the correct way: https://ffmpeg.org/ffmpeg-protocols.html#rtmp
[14:56:35 CEST] <DHE> friki2015: you running this under perl or something that might interpret the '@' symbol then?
[15:14:30 CEST] <friki2015> DHE: nope, i'm using bash like this: ffmpeg -i - ... -f flv 'rtmp://a:b@host/path'
[15:25:22 CEST] <friki2015> fixed :) i've added librtmp to my ffmpeg compilation (build in rtmp seems to fail on auth)
[15:35:03 CEST] <soulshock> is it possible to apply the resize and deinterlace filter once and then create several different bitrates, without having to run the filters for each bitrate?
[15:36:42 CEST] <soulshock> right now this is my cmdline but it is extremely slow (speed=0.04x)
[15:36:42 CEST] <soulshock> -i "8884552094154c739be5a63900a7d42c.mov" -filter_complex "yadif=0:-1:0, scale=1280x720, split=8[out1][out2][out3][out4][out5][out6][out7][out8]" -map [out1] -c:v libx264 -b:v 3000k -an -pix_fmt yuv420p -preset veryslow "1280x720_3000_128_0.mp4" -map [out2] -c:v libx264 -b:v 2750k -an -pix_fmt yuv420p -preset veryslow "1280x720_2750_128_0.mp4" -map [out3] -c:v libx264 -b:v 2500k -an -pix_fmt yuv420p -preset veryslow "1280x720_2500_128_0.mp4" -map [out4]
[15:38:00 CEST] <BtbN> that already does just that. You have 8 x264 on veryslow, what did you expect?
[15:38:26 CEST] <BtbN> could put the pix_fmt conversion in the filter, so it also happens only once. But apart from that.
[15:38:55 CEST] <soulshock> ok
[15:39:02 CEST] <furq> yeah eight 720p veryslow conversions are going to send your cpu to hell
[15:39:06 CEST] <furq> especially with no explicit -threads
[15:39:11 CEST] <bencoh> huhu
[15:39:24 CEST] <soulshock> I want it to use all the cores, so no -threads would be optimal ?
[15:39:27 CEST] <bencoh> cpu and mem too (because of threads)
[15:39:39 CEST] <furq> if you have 8 cores then that command is using 96 threads
[15:39:44 CEST] <bencoh> it would be "optimal" if you had only one encoder running
[15:39:46 CEST] <furq> which doesn't seem great
[15:40:04 CEST] <furq> although i'm not an expert on scheduling
[15:40:23 CEST] <soulshock> ok
[15:40:36 CEST] <furq> in fact more than that because of lookahead threads
[15:40:48 CEST] <BtbN> ffmpeg also isn't exactly efficient with multiple outputs
[15:40:54 CEST] <BtbN> it works through them seqeuntially
[15:41:11 CEST] <bencoh> it's still better than spawning 8 yadif though
[15:41:17 CEST] <bencoh> yadif isn't exactly cheap either :)
[15:41:34 CEST] <furq> also if you're encoding multiple bitrates then i assume this is for streaming
[15:41:42 CEST] <furq> in which case you probably don't want to be using 8 bframes and 16 refs
[15:41:42 CEST] <BtbN> you'd get better results by using an intermediate lossless file, and then 8 independend ffmpeg processes encoding that
[15:42:09 CEST] <soulshock> yeah I'm playing with generating 10 bitrates and then picking the lowest bitrate which is still acceptable quality
[15:42:38 CEST] <furq> i wouldn't use -b:v at all unless you're trying to hit a specific filesize
[15:43:58 CEST] <soulshock> so -crf?
[15:44:03 CEST] <furq> yeah
[15:44:11 CEST] <BtbN> or qp
[15:44:36 CEST] <furq> is qp preferable for anything but lossles
[15:44:37 CEST] <furq> s
[15:44:49 CEST] <soulshock> good point. filesize doesn't really matter, just making sure it doesn't generate insane bitrates so people's adsl at home can still handle it
[15:44:59 CEST] <furq> oh right, you are streaming
[15:45:24 CEST] <soulshock> so bitrate still matters
[15:46:01 CEST] <soulshock> I still need to try 2-pass vbr and crf though
[15:46:45 CEST] <soulshock> ok gotta go. thanks for the input :)
[15:46:49 CEST] <furq> you probably want to specify -level 41
[15:46:50 CEST] <furq> dfs
[15:49:37 CEST] <BtbN> 2-pass vbr and crf?
[15:49:51 CEST] <furq> i assume he meant separately
[16:04:36 CEST] <UnDeRsOuL> hello, videotoolbox framework in OS X it´s only designed to make use of hardware decoding for h264 only?
[16:08:57 CEST] <rkern> UnDeRsOuL: yes, the hwaccel only supports h.264 hardware decoding. The other codecs use software decoding with videotoolbox.
[16:11:58 CEST] <UnDeRsOuL> Who to "blame" in my case? Apple or Intel (which is my videocard)
[16:12:44 CEST] <UnDeRsOuL> Obviously Apple, they are the one who made the framework. :/
[16:17:05 CEST] <rkern> Does your board support hardware decoding for a format that's using software with videotoolbox?
[16:17:15 CEST] <edoardo> Hi!
[16:17:47 CEST] <edoardo> I'd like to slide a spectrogram over a dark background at the exact speed of 32 pixels per seconds
[16:17:52 CEST] <edoardo> And add some music to it
[16:18:10 CEST] <edoardo> ffmpeg -i song.flac -i spectrogram.png -filter "overlay=0:x=1280-t*32:y=0" out.webm
[16:18:30 CEST] <edoardo> This one doesn't work because i never really used compositing on ffmpeg. Any help?
[16:22:58 CEST] <UnDeRsOuL> rkern: so it´s the codec then. Thx. I´ll look for codecs that support hardware decoding then. Thx again
[17:21:39 CEST] <Angus> Hi all
[17:22:12 CEST] <Angus> Trying to build a program in both OpenSUSE and Ubuntu, but it seems OpenSUSE has the libav libraries in /usr/include/ffmpeg instead of just /usr/include
[17:22:37 CEST] <Angus> is there a way to generate a config.h file to account for this?
[17:23:07 CEST] <Angus> or modify #include blocks to account for this?
[17:23:31 CEST] <furq> Angus: pkg-config --cflags libavcodec
[17:25:57 CEST] <Angus> furq: Cheers!
[17:26:51 CEST] <p3terpan> hi there
[17:27:20 CEST] <p3terpan> i would like to know how can i restream with 1 hour delay?
[17:35:41 CEST] <Angus> libavutil/hwcontext.h is part of the public API...?
[18:45:45 CEST] <BtbN> Angus, yes.
[18:52:47 CEST] <squirrel> when writing to a usb stick, ffmpeg seems to be exceptionally slow, slower than writing to a temp folder and then copying result to the stick. can i increase output buffer or something?
[18:57:05 CEST] <BtbN> That's mostly up to your file system.
[18:57:15 CEST] <BtbN> copying large chunks is faster
[19:01:21 CEST] <squirrel> that's what i thought, but doesn't ffmpeg do some kind of buffered output
[19:16:50 CEST] <digriz> anyone have ffmpeg precompiled for windows w/ nvenc enabled
[19:17:17 CEST] <BtbN> Didn't Zeranoe enable that quite a while ago?
[19:17:21 CEST] <soulshock> the latest builds have nvenc yes
[19:17:41 CEST] <digriz> i think it needs some flag set when its compiled tho?
[19:18:09 CEST] <soulshock> yes and it has that
[19:19:15 CEST] <digriz> does it matter if i get shared or static?
[19:22:04 CEST] <digriz> guess ill try static
[19:28:47 CEST] <Fyr> Zeranoe's site doesn't tell anything about NVENC.
[19:33:02 CEST] <Fyr> cool, it has --enable-nvenc.
[19:33:52 CEST] <digriz> i dont think it works
[19:33:56 CEST] <digriz> thats why i had asked
[19:34:19 CEST] <Fyr> -hwaccels shows only dxva2. =/
[19:34:21 CEST] <digriz> ill have to go into the command line later to see its output
[19:34:25 CEST] <digriz> yeah
[19:34:32 CEST] <digriz> its a liscensing issue
[19:34:44 CEST] <digriz> it has to be compiled w/ some flag to put it on when compiled
[19:34:51 CEST] <digriz> then supposedly you cant redistribute it
[19:34:54 CEST] <digriz> which is stupid
[19:35:16 CEST] <digriz> but you can compile it thats allowed
[19:35:26 CEST] <digriz> i dont have tools to do such things on my machine
[19:37:42 CEST] <digriz> i guess i could compile it on a linux box, for windows? i dunno
[19:37:56 CEST] <digriz> https://bannsecurity.com/index.php/tutorials/23-compiling-ffmpeg-for-windows-nvidia
[19:38:06 CEST] <digriz> this article is a bit old-ish
[19:40:29 CEST] <digriz> maybe i need --enable-nonfree
[19:40:57 CEST] <digriz> not sure why its called that since it is actually free
[19:41:19 CEST] <digriz> maybe not free for commercial use?
[19:42:54 CEST] <digriz> https://ffmpeg.zeranoe.com/forum/viewtopic.php?f=7&t=3666&p=10289&hilit=nvenc#p10289
[19:42:57 CEST] <digriz> this person had issues too
[19:44:43 CEST] <digriz> someone asked for the bin files too like me heh
[19:46:19 CEST] <soulshock> weird I tried nvenc and it worked fine a few weeks ago
[19:46:47 CEST] <digriz> maybe somethings changed
[19:46:58 CEST] <digriz> do you still have your files from a few weeks ago :)
[19:47:01 CEST] <digriz> *beg*
[19:47:30 CEST] <digriz> maybe they're archived on zeranoe
[19:47:35 CEST] <MachinaeWolf> Is there a way to record audio with ffmpeg and audacity like you can with alsa?
[19:48:13 CEST] <BtbN> digriz, your information is outdated.
[19:48:15 CEST] <digriz> they are archived it looks like if you go back a directory
[19:48:22 CEST] <BtbN> nvenc binaries can be freely distributed.
[19:48:35 CEST] <BtbN> And nvenc is not a hwaccel.
[19:48:39 CEST] <BtbN> Just another encoder.
[19:54:26 CEST] <BtbN> current zeranoe build has h264_nvenc and hevc_nvenc just fine.
[19:59:12 CEST] <BtbN> digriz, you might have to add -pix_fmt nv12 for nvenc to behave, no idea what's up with that. Will investigate.
[19:59:38 CEST] <pood> hi there
[20:01:03 CEST] <pood> is it possible to make a videoconference with ffserver or so?
[20:10:47 CEST] <digriz> BtbN: thank you
[20:31:34 CEST] <soulshock> does ffmpeg have any hwaccel encoding support? meaning using x264 with hardware accelleration
[20:31:42 CEST] <soulshock> or rather, does any hardware even exist for that
[20:32:29 CEST] <BtbN> x264 does not support hardware acceleration. There is some rudimentary OpenCL support, but it's mostly pointless.
[20:32:48 CEST] <soulshock> ok
[20:33:05 CEST] <BtbN> You'll have to use one of the various hardware encoders, which basically trade CPU usage for encoding quality/efficiency.
[20:33:35 CEST] <soulshock> yeah I've tried nvenc and didn't like the quality. but I guess it has it's place
[20:34:00 CEST] <BtbN> The quality depends a lot on your settings and what you are trying to achive.
[20:34:37 CEST] <soulshock> yea
[23:49:18 CEST] <Threads> just wondering is this right syntax for using weston three field -vf w3fdif=simple:all
[00:00:00 CEST] --- Wed Jul  6 2016


More information about the Ffmpeg-devel-irc mailing list